Re-examining Double Descent and Scaling Laws under Norm-based Capacity via Deterministic Equivalence

• URL: http://arxiv.org/abs/2502.01585v1
• Date: Mon, 03 Feb 2025 18:10:40 GMT
• Title: Re-examining Double Descent and Scaling Laws under Norm-based Capacity via Deterministic Equivalence
• Authors: Yichen Wang, Yudong Chen, Lorenzo Rosasco, Fanghui Liu,
• Abstract summary: We investigate double descent and scaling laws in terms of weights rather than the number of parameters. Our results rigorously answer whether double descent exists under norm-based capacity and reshape the corresponding scaling laws.
• Score: 20.88908358215574
• License:
• Abstract: We investigate double descent and scaling laws in terms of weights rather than the number of parameters. Specifically, we analyze linear and random features models using the deterministic equivalence approach from random matrix theory. We precisely characterize how the weights norm concentrate around deterministic quantities and elucidate the relationship between the expected test error and the norm-based capacity (complexity). Our results rigorously answer whether double descent exists under norm-based capacity and reshape the corresponding scaling laws. Moreover, they prompt a rethinking of the data-parameter paradigm - from under-parameterized to over-parameterized regimes - by shifting the focus to norms (weights) rather than parameter count.

Related papers

• Norm-Bounded Low-Rank Adaptation [10.22454500514559]
  We introduce two parameterizations that allow explicit bounds on each singular value of the weight adaptation matrix. Experiments on vision fine-tuning benchmarks show that the proposed approach can achieve good adaptation performance. We also explore applications in privacy-preserving model merging and low-rank matrix completion.
  arXiv  Detail & Related papers  (2025-01-31T11:24:57Z)
• Scaling and renormalization in high-dimensional regression [72.59731158970894]
  This paper presents a succinct derivation of the training and generalization performance of a variety of high-dimensional ridge regression models. We provide an introduction and review of recent results on these topics, aimed at readers with backgrounds in physics and deep learning.
  arXiv  Detail & Related papers  (2024-05-01T15:59:00Z)
• Variational Bayesian surrogate modelling with application to robust design optimisation [0.9626666671366836]
  Surrogate models provide a quick-to-evaluate approximation to complex computational models. We consider Bayesian inference for constructing statistical surrogates with input uncertainties and dimensionality reduction. We demonstrate intrinsic and robust structural optimisation problems where cost functions depend on a weighted sum of the mean and standard deviation of model outputs.
  arXiv  Detail & Related papers  (2024-04-23T09:22:35Z)
• Gradient-based bilevel optimization for multi-penalty Ridge regression through matrix differential calculus [0.46040036610482665]
  We introduce a gradient-based approach to the problem of linear regression with l2-regularization. We show that our approach outperforms LASSO, Ridge, and Elastic Net regression. The analytical of the gradient proves to be more efficient in terms of computational time compared to automatic differentiation.
  arXiv  Detail & Related papers  (2023-11-23T20:03:51Z)
• Likelihood Ratio Confidence Sets for Sequential Decision Making [51.66638486226482]
  We revisit the likelihood-based inference principle and propose to use likelihood ratios to construct valid confidence sequences. Our method is especially suitable for problems with well-specified likelihoods. We show how to provably choose the best sequence of estimators and shed light on connections to online convex optimization.
  arXiv  Detail & Related papers  (2023-11-08T00:10:21Z)
• Enriching Disentanglement: From Logical Definitions to Quantitative Metrics [59.12308034729482]
  Disentangling the explanatory factors in complex data is a promising approach for data-efficient representation learning. We establish relationships between logical definitions and quantitative metrics to derive theoretically grounded disentanglement metrics. We empirically demonstrate the effectiveness of the proposed metrics by isolating different aspects of disentangled representations.
  arXiv  Detail & Related papers  (2023-05-19T08:22:23Z)
• Post-mortem on a deep learning contest: a Simpson's paradox and the complementary roles of scale metrics versus shape metrics [61.49826776409194]
  We analyze a corpus of models made publicly-available for a contest to predict the generalization accuracy of neural network (NN) models. We identify what amounts to a Simpson's paradox: where "scale" metrics perform well overall but perform poorly on sub partitions of the data. We present two novel shape metrics, one data-independent, and the other data-dependent, which can predict trends in the test accuracy of a series of NNs.
  arXiv  Detail & Related papers  (2021-06-01T19:19:49Z)
• Benign Overfitting of Constant-Stepsize SGD for Linear Regression [122.70478935214128]
  inductive biases are central in preventing overfitting empirically. This work considers this issue in arguably the most basic setting: constant-stepsize SGD for linear regression. We reflect on a number of notable differences between the algorithmic regularization afforded by (unregularized) SGD in comparison to ordinary least squares.
  arXiv  Detail & Related papers  (2021-03-23T17:15:53Z)
• Understanding Implicit Regularization in Over-Parameterized Single Index Model [55.41685740015095]
  We design regularization-free algorithms for the high-dimensional single index model. We provide theoretical guarantees for the induced implicit regularization phenomenon.
  arXiv  Detail & Related papers  (2020-07-16T13:27:47Z)
• Asymptotics of Ridge (less) Regression under General Source Condition [26.618200633139256]
  We consider the role played by the structure of the true regression parameter. We show that (no regularisation) can be optimal even with bounded signal-to-noise ratio (SNR) This contrasts with previous work considering ridge regression with isotropic prior, in which case is only optimal in the limit of infinite SNR.
  arXiv  Detail & Related papers  (2020-06-11T13:00:21Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.